Cloudera: 72% of Korean Companies Are Increasing AI Investment, but Governance and Data Sovereignty Are Hurdles to Expansion

Cloudera released its report, "State of Enterprise AI in 2026: The New Rules of Scaling," on the 4th. The report showed that Korean companies are actively increasing AI investment. The survey covered 1,500 IT leaders and decision-makers across 9 countries, including South Korea, and was conducted jointly with Wakefield Research from Sept. 2 to 14.

According to the report, Korean companies were more aggressive than the global average in expanding AI investment. Seventy-two percent of Korean companies said they were increasing AI investment. The domestic respondents were 100 IT decision-makers.

However, data governance issues, data sovereignty concerns, and a lack of specialized talent were cited as obstacles to expanding AI operations. The domestic survey items included AI investment and economics, expansion challenges in real-world operating environments, the current state of data sovereignty, and infrastructure operations.

Cloudera held an online pre-media briefing on the 3rd ahead of the release. The briefing introduced key survey findings and the hybrid AI platform strategy. The presentation was given by Sergio Gago, Cloudera's CTO.

Seventy-two percent of domestic respondents said AI investment increased over the past year. That figure exceeded the global average of 55% and the Asia-Pacific average of 58%. Nineteen percent of domestic respondents said AI investment increased significantly.

Investment levels were also high. Sixty-eight percent of domestic respondents said they spend more than USD 500,000 a year on AI projects. Forty-nine percent of domestic respondents said they invest more than USD 1 million a year in AI projects.

The share investing more than USD 1 million a year came to 28% globally and 32% in the Asia-Pacific region. As a result, the share of domestic respondents investing more than USD 1 million a year was higher than the global and Asia-Pacific averages.

However, constraints were also identified at the stage of connecting AI investment to actual work and service expansion. Sixty-two percent of domestic respondents cited governance and compliance as a barrier to AI scaling, and 56% pointed to data sovereignty. This was followed by 51% citing a lack of technical expertise, 50% citing data integration, and 40% citing token costs and AI inference.

Seung-chul Choi, head of Cloudera Korea, said Korean companies are rapidly expanding AI investment. He explained that governance and data sovereignty are emerging as constraints to AI scaling amid pressure from rising costs. He added that more than half of domestic respondents identified data sovereignty as a barrier to AI scaling, and said a hybrid strategy that allows AI to run in the desired environment under consistent governance without moving data is essential.

Global companies appear to be adjusting how they use AI depending on operating costs. Eighty-nine percent of respondents said they had limited or redesigned AI use cases because of rising token usage. In addition, 45% said they had moved AI workloads in response to rising costs and greater volatility, with destinations including other clouds, private environments, and on-premises systems.

Operational stability in real-world settings also emerged as an issue. Ninety-eight percent of global respondents said they had rolled back an AI deployment at least once because of outages, hallucinations, or unexpected behavior. Meanwhile, 43% said productivity gains exceeded expectations, while 23% said they fell short.

Brian Martin, CTO of Applied AI at Cloudera, said the current effective way of operating AI could prove unsuitable in terms of cost and efficiency in the future. He explained that the execution environment suitable for a workload also changes depending on cost, model, and regulatory shifts.

Against this backdrop of possible change, where AI workloads are operated has emerged as a key issue. Eighty-two percent of domestic respondents said data storage location, data sovereignty, and country-specific regulatory requirements affect workload placement. This is similar to the global average of 84%.

These factors were also cited as reasons Korean companies operate AI across multiple environments. The top reason Korean companies run AI across multiple environments was data sovereignty and residency requirements, at 47%. Another 42% said they do so to keep AI workloads close to enterprise data.

However, changing operating environments was found to entail a significant burden. Sixty-six percent of domestic respondents said that moving AI workloads to another environment requires at least a moderate redesign.

That figure is 8 percentage points higher than the global average of 58%. Even as the suitable execution environment changes in line with cost, model, and regulatory shifts, Korean companies appear to feel a greater redesign burden when actually switching environments.

At the pre-briefing on the 3rd, CTO Gago said that with the introduction of AI-related laws and regulations in Korea and other countries, companies must review how to manage AI governance and how to protect data privacy before adopting technology.

As a response, Gago introduced an approach that sends AI to where the data is, saying it helps address related issues. He then proposed connecting AI to existing data lakes and databases as an alternative to moving data separately for AI, saying this would allow existing data policies and governance to be maintained.

He also said the importance of data location and management methods is increasing as AI agents expand distributed data access across multiple environments. He stressed that as companies broaden AI use, they must also consider token-usage-based costs and establish operating strategies that can secure ROI.

Cloudera introduced "Cloudera Anywhere Cloud" as a platform to address enterprise AI and data operations challenges. The platform is a modular hybrid data and AI platform designed to support AI and data operations.

"Cloudera Anywhere Cloud" targets multiple environments, including public clouds and on-premises systems, and is built on the premise of leveraging existing data lakes. It also supports running AI workloads in the environment where the data resides, rather than moving the data.

Cloudera said the platform is focused on applying consistent security regardless of environment, and also on applying consistent governance regardless of environment. It added that the platform enables companies to choose operating environments according to their cost, performance, and regulatory requirements.

CTO Gago said Korean companies are having difficulty with AI governance and heterogeneous infrastructure management. He identified dependence on U.S. companies for the latest AI models as a major challenge, and also cited securing GPU and a shortage of specialized talent as major issues.

Although constraints exist, AI investment and demand among Korean companies remain at a higher level than in other countries, and the outlook for the domestic market was presented as positive despite these constraints.

Accordingly, Cloudera said it plans to support domestic customers around Anywhere Cloud and help them expand the scope of AI adoption while leveraging existing data infrastructure. Choi said AI's role in enterprise decision-making is becoming more significant, and that Cloudera aims to provide domestic customers with the flexibility and control needed for AI scaling to support continued growth.

Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=409234

References

• https://www.epnc.co.kr/news/articleView.html?idxno=409234